The present technology relates to a power demand forecast device, method and system and power failure detection system.
In a number of nations, supply of power to power consumers has been left exclusively in the hands of electric utilities. In recent years, however, entry of entities other than electric utilities into power retailing business and power futures trading have been realized or promoted in a bid to introduce competition into the electric power industry and supply power at lower prices.
Power futures trading refers, for example, to forecasting the necessary amount of electric power in advance and selling or buying electric power up to the following day or 24 hours later through the electricity market. In order for an entity engaged in power retailing and power futures trading as its line of business to hold superiority in trading and make large profits, it is necessary for that entity to make an accurate forecast of power demand.
As a technique used to forecast power demand, a total power demand forecast device has been proposed that forecasts total power demand by accepting meteorological variables including past temperatures and humidities and total power demand data and further learning in neural networks (see Japanese Patent Laid-Open No. Hei 5-18995, hereinafter referred to as Patent Document 1).